Regression Check

Re-run the baseline suite and compare against the last known-good run. Flags metric drift, test failures, or behavioral changes.

swarm-ai-safety Updated

File contents

EXECUTE NOW

Mode: $ARGUMENTS

Parse:

  • tests — run pytest suite only
  • baseline — re-run baseline scenario and compare metrics
  • full or empty — run both tests and baseline comparison

Step 1: Run tests

python -m pytest tests/ -v --tb=short

Report pass/fail count. If any failures, list them.

Step 2: Run baseline scenario

python -m swarm run scenarios/baseline.yaml --seed 42 --epochs 10 --steps 10

Step 3: Compare against last known-good

Read the most recent baseline run from memory at .letta/memory/runs/latest.md.

If a previous baseline exists, compare:

  • Welfare: delta and percentage change
  • Toxicity: delta
  • Acceptance rate: delta
  • Quality gap: delta

Flag any metric that changed by more than 10% as a potential regression.

Step 4: Report

## Regression Check

### Tests
- {passed}/{total} passed
- Failures: {list or "none"}

### Baseline comparison
| Metric | Previous | Current | Delta | Status |
|--------|----------|---------|-------|--------|
| Welfare | ... | ... | ... | OK/DRIFT |
| Toxicity | ... | ... | ... | OK/DRIFT |

### Verdict
{CLEAN — no regressions detected}
{REGRESSED — {details}}

swarm-ai-safety/swarm/tree/main/.skills/regression-check commit 37a3f67270

Frequently asked questions

npx skillmds@latest add swarm-ai-safety/regression-check